Forgetting: A New Mechanism Towards Better Large Language Model Fine-tuning

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Taheri, Ali, Taban, Alireza, Wang, Qizhou, Ye, Shanshan, Mirzaei, Abdolreza, Liu, Tongliang, Han, Bo
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866914428215099392
author Taheri, Ali
Taban, Alireza
Wang, Qizhou
Ye, Shanshan
Mirzaei, Abdolreza
Liu, Tongliang
Han, Bo
author_facet Taheri, Ali
Taban, Alireza
Wang, Qizhou
Ye, Shanshan
Mirzaei, Abdolreza
Liu, Tongliang
Han, Bo
contents Supervised fine-tuning (SFT) plays a critical role for pretrained large language models (LLMs), notably enhancing their capacity to acquire domain-specific knowledge while preserving or potentially augmenting their general-purpose capabilities. However, the efficacy of SFT hinges on data quality as well as data volume, otherwise it may result in limited performance gains or even degradation relative to the associated baselines. To mitigate such reliance, we suggest categorizing tokens within each corpus into two parts -- positive and negative tokens -- based on whether they are useful to improve model performance. Positive tokens can be trained in common ways, whereas negative tokens, which may lack essential semantics or be misleading, should be explicitly forgotten. Overall, the token categorization facilitates the model to learn less informative messages, and the forgetting guides the model on what information to learn more precisely. We conduct experiments across diverse and well-established benchmarks using various model architectures, demonstrating that this forgetting mechanism enhances model performance.
format Preprint
id arxiv_https___arxiv_org_abs_2508_04329
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Forgetting: A New Mechanism Towards Better Large Language Model Fine-tuning
Taheri, Ali
Taban, Alireza
Wang, Qizhou
Ye, Shanshan
Mirzaei, Abdolreza
Liu, Tongliang
Han, Bo
Machine Learning
Supervised fine-tuning (SFT) plays a critical role for pretrained large language models (LLMs), notably enhancing their capacity to acquire domain-specific knowledge while preserving or potentially augmenting their general-purpose capabilities. However, the efficacy of SFT hinges on data quality as well as data volume, otherwise it may result in limited performance gains or even degradation relative to the associated baselines. To mitigate such reliance, we suggest categorizing tokens within each corpus into two parts -- positive and negative tokens -- based on whether they are useful to improve model performance. Positive tokens can be trained in common ways, whereas negative tokens, which may lack essential semantics or be misleading, should be explicitly forgotten. Overall, the token categorization facilitates the model to learn less informative messages, and the forgetting guides the model on what information to learn more precisely. We conduct experiments across diverse and well-established benchmarks using various model architectures, demonstrating that this forgetting mechanism enhances model performance.
title Forgetting: A New Mechanism Towards Better Large Language Model Fine-tuning
topic Machine Learning
url https://arxiv.org/abs/2508.04329